REACT3D: Real-time Edge Accelerator for Incremental Training in 3D Gaussian Splatting based SLAM Systems

3D Gaussian Splatting (3DGS) has emerged as a promising approach for high-fidelity scene reconstruction and has been widely adopted in Simultaneous Localization and Mapping (SLAM) systems. 3DGS SLAM requires incremental training and rendering of Gaussians in real-time from continuous camera viewpoints. To match the streaming nature of SLAM, 3DGS-based mapping must sustain over 30 frames per second (FPS), which is a widely recognized threshold for maintaining accurate tracking and mapping quality. Existing GPU-based solutions and prior accelerators fall short of this target, primarily due to redundant training computation, unnecessary loss computing, and irregular memory access patterns. To address these challenges, we propose REACT3D, a real-time edge accelerator designed for incremental training in 3DGS SLAM systems. At the algorithmic level, we introduce spatial consistency and convergence aware sparsification, which eliminates redundant computation in both forward and backward rendering by predicting under-optimized regions based on spatial coherence and convergence dynamics. At the architectural level, we design a pixel block-wise fine-grained dataflow to eliminate explicit loss computing, establish a tightly coupled pipeline, and improve hardware utilization. Furthermore, we develop a Content Addressable Memory (CAM)-based Dual-index Gaussian Buffer to resolve discontinuous memory accesses and avoid computation conflicts during gradient merging. Experimental results show that REACT3D achieves 12.10 × performance and 33.69 × energy efficiency improvement over NVIDIA Jetson AGX Orin. To the best of our knowledge, REACT3D is the first hardware design that meets the real-time requirements of 3DGS SLAM on the edge.